diff --git a/deploy/lite/Makefile_armv7 b/deploy/lite/Makefile_armv7
new file mode 100644
index 0000000000000000000000000000000000000000..2d412a28890d32bab781446ff3ab3369d120a830
--- /dev/null
+++ b/deploy/lite/Makefile_armv7
@@ -0,0 +1,61 @@
+ARM_ABI = arm7
+export ARM_ABI
+
+include ../Makefile.def
+
+LITE_ROOT=../../../
+
+THIRD_PARTY_DIR=${LITE_ROOT}/third_party
+
+OPENCV_VERSION=opencv4.1.0
+
+OPENCV_LIBS = ${THIRD_PARTY_DIR}/${OPENCV_VERSION}/armeabi-v7a/libs/libopencv_imgcodecs.a \
+              ${THIRD_PARTY_DIR}/${OPENCV_VERSION}/armeabi-v7a/libs/libopencv_imgproc.a \
+              ${THIRD_PARTY_DIR}/${OPENCV_VERSION}/armeabi-v7a/libs/libopencv_core.a \
+              ${THIRD_PARTY_DIR}/${OPENCV_VERSION}/armeabi-v7a/3rdparty/libs/libtegra_hal.a \
+              ${THIRD_PARTY_DIR}/${OPENCV_VERSION}/armeabi-v7a/3rdparty/libs/liblibjpeg-turbo.a \
+              ${THIRD_PARTY_DIR}/${OPENCV_VERSION}/armeabi-v7a/3rdparty/libs/liblibwebp.a \
+              ${THIRD_PARTY_DIR}/${OPENCV_VERSION}/armeabi-v7a/3rdparty/libs/liblibpng.a \
+              ${THIRD_PARTY_DIR}/${OPENCV_VERSION}/armeabi-v7a/3rdparty/libs/liblibjasper.a \
+              ${THIRD_PARTY_DIR}/${OPENCV_VERSION}/armeabi-v7a/3rdparty/libs/liblibtiff.a \
+              ${THIRD_PARTY_DIR}/${OPENCV_VERSION}/armeabi-v7a/3rdparty/libs/libIlmImf.a \
+              ${THIRD_PARTY_DIR}/${OPENCV_VERSION}/armeabi-v7a/3rdparty/libs/libtbb.a \
+              ${THIRD_PARTY_DIR}/${OPENCV_VERSION}/armeabi-v7a/3rdparty/libs/libcpufeatures.a
+
+OPENCV_INCLUDE = -I../../../third_party/${OPENCV_VERSION}/armeabi-v7a/include
+
+CXX_INCLUDES = $(INCLUDES) ${OPENCV_INCLUDE} -I$(LITE_ROOT)/cxx/include
+
+CXX_LIBS = ${OPENCV_LIBS} -L$(LITE_ROOT)/cxx/lib/ -lpaddle_light_api_shared $(SYSTEM_LIBS)
+
+###############################################################
+# How to use one of static libaray:                           #
+#  `libpaddle_api_full_bundled.a`                             #
+#  `libpaddle_api_light_bundled.a`                            #
+###############################################################
+# Note: default use lite's shared library.                    #
+###############################################################
+# 1. Comment above line using `libpaddle_light_api_shared.so`
+# 2. Undo comment below line using `libpaddle_api_light_bundled.a`
+
+#CXX_LIBS = $(LITE_ROOT)/cxx/lib/libpaddle_api_light_bundled.a $(SYSTEM_LIBS)
+
+detect_system: fetch_opencv detect_system.o
+	$(CC) $(SYSROOT_LINK) $(CXXFLAGS_LINK) detect_system.o -o detect_system  $(CXX_LIBS) $(LDFLAGS)
+
+detect_system.o: run_detection.cc
+	$(CC) $(SYSROOT_COMPLILE) $(CXX_DEFINES) $(CXX_INCLUDES) $(CXX_FLAGS) -o detect_system.o -c run_detection.cc
+
+fetch_opencv:
+	@ test -d ${THIRD_PARTY_DIR} ||  mkdir ${THIRD_PARTY_DIR}
+	@ test -e ${THIRD_PARTY_DIR}/${OPENCV_VERSION}.tar.gz || \
+      (echo "fetch opencv libs" && \
+      wget -P ${THIRD_PARTY_DIR} https://paddle-inference-dist.bj.bcebos.com/${OPENCV_VERSION}.tar.gz)
+	@ test -d ${THIRD_PARTY_DIR}/${OPENCV_VERSION} || \
+      tar -zxvf ${THIRD_PARTY_DIR}/${OPENCV_VERSION}.tar.gz -C ${THIRD_PARTY_DIR}
+
+
+.PHONY: clean
+clean:
+	rm -f detect_system.o
+	rm -f detect_system
diff --git a/deploy/lite/README.md b/deploy/lite/README.md
index d32fd5434af885069313573b8ec1af9ab2c649ca..cded48cb5261a77696e2ee6a7ac847e275b2112d 100644
--- a/deploy/lite/README.md
+++ b/deploy/lite/README.md
@@ -12,7 +12,7 @@ Paddle Lite是飞桨轻量化推理引擎，为手机、IOT端提供高效推理
 
 ### 1.1 准备交叉编译环境
 交叉编译环境用于编译 Paddle Lite 和 PaddleDetection 的C++ demo。
-支持多种开发环境，不同开发环境的编译流程请参考对应文档。
+支持多种开发环境，不同开发环境的编译流程请参考对应文档，请确保安装完成Java jdk、Android NDK(R17以上)。
 
 1. [Docker](https://paddle-lite.readthedocs.io/zh/latest/source_compile/compile_env.html#docker)
 2. [Linux](https://paddle-lite.readthedocs.io/zh/latest/source_compile/compile_env.html#linux)
@@ -25,9 +25,9 @@ Paddle Lite是飞桨轻量化推理引擎，为手机、IOT端提供高效推理
       |平台|预测库下载链接|
       |-|-|
       |Android|[arm7](https://github.com/PaddlePaddle/Paddle-Lite/releases/download/v2.8/inference_lite_lib.android.armv7.gcc.c++_static.with_extra.with_cv.tar.gz) / [arm8](https://github.com/PaddlePaddle/Paddle-Lite/releases/download/v2.8/inference_lite_lib.android.armv8.gcc.c++_static.with_extra.with_cv.tar.gz)|
-      |iOS|[arm7](https://github.com/PaddlePaddle/Paddle-Lite/releases/download/v2.8/inference_lite_lib.ios.armv7.with_cv.with_extra.tiny_publish.tar.gz) / [arm8](https://github.com/PaddlePaddle/Paddle-Lite/releases/download/v2.8/inference_lite_lib.ios.armv8.with_cv.with_extra.tiny_publish.tar.gz)|
 
-      注：1. 如果是从 Paddle-Lite [官方文档](https://paddle-lite.readthedocs.io/zh/latest/quick_start/release_lib.html#android-toolchain-gcc)下载的预测库，注意选择`with_extra=ON，with_cv=ON`的下载链接。
+**注意**：1. 如果是从 Paddle-Lite [官方文档](https://paddle-lite.readthedocs.io/zh/latest/quick_start/release_lib.html#android-toolchain-gcc)下载的预测库，注意选择`with_extra=ON，with_cv=ON`的下载链接。2. 目前只提供Android端demo，IOS端demo可以参考[Paddle-Lite IOS demo](https://github.com/PaddlePaddle/Paddle-Lite-Demo/tree/master/PaddleLite-ios-demo)
+
 
 2. 编译Paddle-Lite得到预测库，Paddle-Lite的编译方式如下：
 ```shell
@@ -228,6 +228,7 @@ precision fp32                      # 模型精度
 
 ```shell
 # 执行编译，得到可执行文件detect_system
+# 如果是编译armv7的执行程序，需要使用 Makefile_armv7 替换 Makefile 文件
 make
 
 # 将编译得到的可执行文件移动到debug文件夹中
diff --git a/deploy/lite/run_detection.cc b/deploy/lite/run_detection.cc
index 441d45cec5feac9f5f36991492405664723d21b5..8a84a59db04e482f04012f34e3c19397f9040cb0 100644
--- a/deploy/lite/run_detection.cc
+++ b/deploy/lite/run_detection.cc
@@ -35,7 +35,7 @@ struct Object {
 // Object for storing all preprocessed data
 struct ImageBlob {
   // image width and height
-  std::vector<float> im_shape_;
+  std::vector<int> im_shape_;
   // Buffer for image data after preprocessing
   const float* im_data_;
   // Scale factor for image size to origin image size
@@ -270,8 +270,8 @@ ImageBlob prepare_imgdata(const cv::Mat& img,
   int width = target_size_[0];
   int height = target_size_[1];
   img_data.im_shape_ = {
-      static_cast<float>(target_size_[0]),
-      static_cast<float>(target_size_[1])
+      static_cast<int>(target_size_[0]),
+      static_cast<int>(target_size_[1])
   };
       
   img_data.scale_factor_ = {
diff --git a/static/deploy/lite/Makefile_armv7 b/static/deploy/lite/Makefile_armv7
new file mode 100644
index 0000000000000000000000000000000000000000..2d412a28890d32bab781446ff3ab3369d120a830
--- /dev/null
+++ b/static/deploy/lite/Makefile_armv7
@@ -0,0 +1,61 @@
+ARM_ABI = arm7
+export ARM_ABI
+
+include ../Makefile.def
+
+LITE_ROOT=../../../
+
+THIRD_PARTY_DIR=${LITE_ROOT}/third_party
+
+OPENCV_VERSION=opencv4.1.0
+
+OPENCV_LIBS = ${THIRD_PARTY_DIR}/${OPENCV_VERSION}/armeabi-v7a/libs/libopencv_imgcodecs.a \
+              ${THIRD_PARTY_DIR}/${OPENCV_VERSION}/armeabi-v7a/libs/libopencv_imgproc.a \
+              ${THIRD_PARTY_DIR}/${OPENCV_VERSION}/armeabi-v7a/libs/libopencv_core.a \
+              ${THIRD_PARTY_DIR}/${OPENCV_VERSION}/armeabi-v7a/3rdparty/libs/libtegra_hal.a \
+              ${THIRD_PARTY_DIR}/${OPENCV_VERSION}/armeabi-v7a/3rdparty/libs/liblibjpeg-turbo.a \
+              ${THIRD_PARTY_DIR}/${OPENCV_VERSION}/armeabi-v7a/3rdparty/libs/liblibwebp.a \
+              ${THIRD_PARTY_DIR}/${OPENCV_VERSION}/armeabi-v7a/3rdparty/libs/liblibpng.a \
+              ${THIRD_PARTY_DIR}/${OPENCV_VERSION}/armeabi-v7a/3rdparty/libs/liblibjasper.a \
+              ${THIRD_PARTY_DIR}/${OPENCV_VERSION}/armeabi-v7a/3rdparty/libs/liblibtiff.a \
+              ${THIRD_PARTY_DIR}/${OPENCV_VERSION}/armeabi-v7a/3rdparty/libs/libIlmImf.a \
+              ${THIRD_PARTY_DIR}/${OPENCV_VERSION}/armeabi-v7a/3rdparty/libs/libtbb.a \
+              ${THIRD_PARTY_DIR}/${OPENCV_VERSION}/armeabi-v7a/3rdparty/libs/libcpufeatures.a
+
+OPENCV_INCLUDE = -I../../../third_party/${OPENCV_VERSION}/armeabi-v7a/include
+
+CXX_INCLUDES = $(INCLUDES) ${OPENCV_INCLUDE} -I$(LITE_ROOT)/cxx/include
+
+CXX_LIBS = ${OPENCV_LIBS} -L$(LITE_ROOT)/cxx/lib/ -lpaddle_light_api_shared $(SYSTEM_LIBS)
+
+###############################################################
+# How to use one of static libaray:                           #
+#  `libpaddle_api_full_bundled.a`                             #
+#  `libpaddle_api_light_bundled.a`                            #
+###############################################################
+# Note: default use lite's shared library.                    #
+###############################################################
+# 1. Comment above line using `libpaddle_light_api_shared.so`
+# 2. Undo comment below line using `libpaddle_api_light_bundled.a`
+
+#CXX_LIBS = $(LITE_ROOT)/cxx/lib/libpaddle_api_light_bundled.a $(SYSTEM_LIBS)
+
+detect_system: fetch_opencv detect_system.o
+	$(CC) $(SYSROOT_LINK) $(CXXFLAGS_LINK) detect_system.o -o detect_system  $(CXX_LIBS) $(LDFLAGS)
+
+detect_system.o: run_detection.cc
+	$(CC) $(SYSROOT_COMPLILE) $(CXX_DEFINES) $(CXX_INCLUDES) $(CXX_FLAGS) -o detect_system.o -c run_detection.cc
+
+fetch_opencv:
+	@ test -d ${THIRD_PARTY_DIR} ||  mkdir ${THIRD_PARTY_DIR}
+	@ test -e ${THIRD_PARTY_DIR}/${OPENCV_VERSION}.tar.gz || \
+      (echo "fetch opencv libs" && \
+      wget -P ${THIRD_PARTY_DIR} https://paddle-inference-dist.bj.bcebos.com/${OPENCV_VERSION}.tar.gz)
+	@ test -d ${THIRD_PARTY_DIR}/${OPENCV_VERSION} || \
+      tar -zxvf ${THIRD_PARTY_DIR}/${OPENCV_VERSION}.tar.gz -C ${THIRD_PARTY_DIR}
+
+
+.PHONY: clean
+clean:
+	rm -f detect_system.o
+	rm -f detect_system
diff --git a/static/deploy/lite/README.md b/static/deploy/lite/README.md
index 2a6b58814286944f2f96dc376d3e0be62b29d804..843515aad814de2519b894db154a9f77c28c117a 100644
--- a/static/deploy/lite/README.md
+++ b/static/deploy/lite/README.md
@@ -12,7 +12,7 @@ Paddle Lite是飞桨轻量化推理引擎，为手机、IOT端提供高效推理
 
 ### 1.1 准备交叉编译环境
 交叉编译环境用于编译 Paddle Lite 和 PaddleDetection 的C++ demo。
-支持多种开发环境，不同开发环境的编译流程请参考对应文档。
+支持多种开发环境，不同开发环境的编译流程请参考对应文档，请确保安装完成Java jdk、Android NDK(R17以上)。
 
 1. [Docker](https://paddle-lite.readthedocs.io/zh/latest/source_compile/compile_env.html#docker)
 2. [Linux](https://paddle-lite.readthedocs.io/zh/latest/source_compile/compile_env.html#linux)
@@ -25,9 +25,9 @@ Paddle Lite是飞桨轻量化推理引擎，为手机、IOT端提供高效推理
       |平台|预测库下载链接|
       |-|-|
       |Android|[arm7](https://github.com/PaddlePaddle/Paddle-Lite/releases/download/v2.8/inference_lite_lib.android.armv7.gcc.c++_static.with_extra.with_cv.tar.gz) / [arm8](https://github.com/PaddlePaddle/Paddle-Lite/releases/download/v2.8/inference_lite_lib.android.armv8.gcc.c++_static.with_extra.with_cv.tar.gz)|
-      |iOS|[arm7](https://github.com/PaddlePaddle/Paddle-Lite/releases/download/v2.8/inference_lite_lib.ios.armv7.with_cv.with_extra.tiny_publish.tar.gz) / [arm8](https://github.com/PaddlePaddle/Paddle-Lite/releases/download/v2.8/inference_lite_lib.ios.armv8.with_cv.with_extra.tiny_publish.tar.gz)|
 
-      注：1. 如果是从 Paddle-Lite [官方文档](https://paddle-lite.readthedocs.io/zh/latest/quick_start/release_lib.html#android-toolchain-gcc)下载的预测库，注意选择`with_extra=ON，with_cv=ON`的下载链接。
+**注意**：1. 如果是从 Paddle-Lite [官方文档](https://paddle-lite.readthedocs.io/zh/latest/quick_start/release_lib.html#android-toolchain-gcc)下载的预测库，注意选择`with_extra=ON，with_cv=ON`的下载链接。2. 目前只提供Android端demo，IOS端demo可以参考[Paddle-Lite IOS demo](https://github.com/PaddlePaddle/Paddle-Lite-Demo/tree/master/PaddleLite-ios-demo)
+
 
 2. 编译Paddle-Lite得到预测库，Paddle-Lite的编译方式如下：
 ```shell
@@ -230,6 +230,7 @@ precision fp32                      # 模型精度
 
 ```shell
 # 执行编译，得到可执行文件detect_system
+# 如果是编译armv7的执行程序，需要使用 Makefile_armv7 替换 Makefile 文件
 make
 
 # 将编译得到的可执行文件移动到debug文件夹中
diff --git a/static/deploy/lite/run_detection.cc b/static/deploy/lite/run_detection.cc
index 15e0d227d95cccc749ef15d14fec1297239a3f63..109cc2fff8ae99dc1554a65afd74c16644e736e2 100644
--- a/static/deploy/lite/run_detection.cc
+++ b/static/deploy/lite/run_detection.cc
@@ -35,7 +35,7 @@ struct Object {
 // Object for storing all preprocessed data
 struct ImageBlob {
   // image width and height
-  std::vector<float> im_shape_;
+  std::vector<int> im_shape_;
   // Buffer for image data after preprocessing
   const float* im_data_;
   std::vector<float> mean_;
@@ -268,8 +268,8 @@ ImageBlob prepare_imgdata(const cv::Mat& img,
   int width = target_size_[0];
   int height = target_size_[1];
   img_data.im_shape_ = {
-      static_cast<float>(target_size_[0]),
-      static_cast<float>(target_size_[1])
+      static_cast<int>(target_size_[0]),
+      static_cast<int>(target_size_[1])
   };
 
   std::vector<float> mean_;